Open Research Online
The Open University’s repository of research publications
and other research outputs
Analysis of Single Circulating Tumor Cells (CTCs) to
Infer Phenotype and Genome Changes in Response to
Therapeutic Pressures in Biliary Tract Cancer
Thesis
How to cite:
Reduzzi, Carolina (2020). Analysis of Single Circulating Tumor Cells (CTCs) to Infer Phenotype and Genome Changes in Response to Therapeutic Pressures in Biliary Tract Cancer. PhD thesis The Open University.
For guidance on citations see FAQs.
c
2020 The Author Version: Version of Record
Copyright and Moral Rights for the articles on this site are retained by the individual authors and/or other copyright owners. For more information on Open Research Online’s data policy on reuse of materials please consult the policies page.
The Open University
Milton Keynes, United Kingdom
Affiliated Research Centre Fondazione IRCCS Istituto Nazionale dei Tumori
Milano, Italia
Analysis of single circulating tumor cells
(CTCs) to infer phenotype and genome
changes in response to therapeutic pressures
in biliary tract cancer
Thesis presented for the Degree of Doctor of Philosophy
The Open University, Milton Keynes (UK)
School of Life, Health and Chemical Sciences
Carolina Reduzzi
M.Sc. in Biology Applied to Biomedical Research
Abstract Biliary tract cancer (BTC) is a highly fatal disease mainly treated with standard chemotherapy, albeit with limited efficacy. New therapeutic options are greatly needed, but the use of targeted treatments is often prevented by the impossibility of obtaining tissue biopsies for molecular characterization. Circulating tumor cells (CTCs) could represent a valuable alternative to tissue biopsies, also allowing a real-time monitoring of disease evolution and therapeutic resistance. Unfortunately, by using conventional CTC detection methods, which are based on the expression of epithelial markers, CTCs have been detected in only a small proportion of BTC patients, thus limiting their implementation in BTC clinical management.
Here, we developed a protocol for the identification of not only the classic epithelial CTCs (eCTCs), but also of non-conventional CTCs (ncCTCs) lacking epithelial and leukocyte markers, but presenting aberrant genomes, and therefore representing bona fide CTCs. CTCs were analyzed in 41 blood samples longitudinally collected from 21 patients with BTC. The detection of ncCTCs in addition to eCTCs resulted in an increase in CTC-positivity from 19% to 83%. The presence of at least 1 eCTC/10 mL of blood at baseline was associated with a significantly shorter median disease-specific survival (9 months vs. 17 months, p=0.03). Conversely, ncCTCs were not prognostic but variations in their number during treatment mirrored patient response, supporting their role for treatment monitoring.
The developed workflow also allowed the molecular characterization of single-CTCs. Copy number alteration profiling was performed for 88 single-CTCs collected from 23 BTC patients. Unsupervised clustering analysis revealed a segregation of CTCs according to patient best response and allowed the identification of genomic regions possibly involved in mechanisms of therapeutic resistance.
Overall, our results demonstrate the presence of a novel subpopulation of CTCs in BTC, paving the way for the use of liquid biopsy to improve clinical management of BTC patients.
This study was conducted under the supervision of Drs. Vera Cappelletti and Nadia Zaffaroni at the Biomarkers Unit, headed by Dr. Maria Grazia Daidone, at the Department of Applied Research & Technological Development, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy.
Table of contents
ABSTRACT ... 3
TABLE OF CONTENTS ... 7
LIST OF FIGURES AND TABLES ... 10
1. INTRODUCTION ... 17
1.1. Biliary tract cancer ... 18
1.1.1. Clinical management of BTC: State of the art ... 20
1.1.2. Clinical management of BTC: Future challenges ... 23
1.2. Circulating tumor cells ... 29
1.2.1. First generation of CTC studies: CTC enumeration ... 31
1.2.1.1.CTC enumeration in patients with BTC ... 34
1.2.2. Technologies to capture CTC heterogeneity ... 35
1.2.3. New generation of CTC studies: CTC characterization at the single-cell level 43 2. SCOPE OF THE THESIS ... 52
3. MATERIALS AND METHODS ... 55
3.1. Patient information and clinical sample collection ... 56
3.2. Cell lines and culture conditions ... 57
3.3. Flow cytometry analysis ... 58
3.4. Spike-in procedure ... 59
3.5. Enrichment methods... 59
3.5.1. ScreenCell® filters® ... 59
3.5.2. AutoMACS® Pro separator ... 60
3.5.3. OncoQuick® ... 60
3.5.4. Parsortix® ... 60
3.6. Post-enrichment procedures ... 62
3.6.1. Fixation and staining ... 62
3.6.2. DEPArrayTM analysis ... 63
3.7. Molecular analysis ... 65
3.7.1. Whole-genome amplification and quality control assay ... 65
3.7.2. Mutational profiling ... 65
3.7.3. Copy number alteration profiling ... 66
3.8. RNA analysis ... 67
4.1.1. Analytical validity assessment ... 73
4.1.2. Enrichment methods ... 80
4.1.2.1.ScreenCell® filters ... 81
4.1.2.2.AutoMACS® Pro separator ... 81
4.1.2.3.OncoQuick® ... 82
4.1.2.4.Parsortix® ... 84
4.1.3. Identification strategies and phenotypic evaluation... 86
4.1.3.1.Positive selection ... 86
4.1.3.2.Negative selection ... 87
4.1.3.3.Phenotypic evaluation ... 88
4.1.4. Whole-genome amplification and quality control assay ... 90
4.1.5. Copy number alteration profiling ... 91
4.2. Detection of CTCs in blood samples from patients with BTC ... 93
4.2.1. Detection of eCTCs and ncCTCs ... 94
4.2.2. Prognostic role of CTCs ... 97
4.2.3. Treatment monitoring ... 99
4.2.4. Phenotypic characterization ... 102
4.2.5. Molecular characterization ... 105
4.2.5.1.Mutational profiling ... 105
4.2.5.2.Copy number alteration analysis ... 107
4.2.5.2.1.Chromosomal instability evaluation ... 107
4.2.5.2.2.Investigation of alterations involved in therapeutic resistance ... 110
4.2.5.3.Double-negative cells in healthy donors ... 113
4.3. Exploratory studies ... 114
4.3.1. Analysis of CTCs’ RNA ... 114
4.3.2. Characterization of dual-positive cells ... 118
5. DISCUSSION ... 122
6. CONCLUSIONS & FUTURE PERSPECTIVES ... 134
7. REFERENCES... 137
PUBLICATIONS ... 166
List of figures and tables
Figure 1.1. Anatomical sub-variants of BTC. ... 18
Figure 1.2. Algorithm for the management of patients with BTC. ... 20
Figure 1.3. Molecular spectrum of intrahepatic and extrahepatic CCA. ... 24
Figure 1.4. Polyclonal seeding and tumor evolution in BTC. ... 26
Figure 1.5. Changes occurring in the molecular landscape of advanced BTC. ... 27
Figure 1.6. CTCs as a real-time liquid biopsy.... 30
Figure 1.7. Clinical validity of CTC enumeration by CellSearch®. ... 32
Figure 1.8. Prognostic impact of CTCs in patients with BTC. ... 34
Figure 1.9. Phenotypic changes in CTCs in response to treatment in breast cancer patients. ... 36
Figure 1.10. New methods for CTC detection. ... 38
Figure 1.11. Standard workflow for single-CTC sequencing. ... 39
Figure 1.12. Genomic alterations in single CTCs. ... 42
Figure 1.13. Dendrogram representing hierarchical clustering and the timeline of sample acquisition. ... 44
Figure 1.14. Intra- and inter-patient genomic heterogeneity of single CTCs. ... 46
Figure 1.15. Clonality and genomic alterations in single CTCs. ... 48
Figure 1.16. CTC CNA-based classifier and clinical outcome. ... 50
Figure 3. 1. Evaluation of DPcells by CellSearch®. ... 69
Figure 4.1. Workflow for CTC analysis. ... 72
Figure 4.2. Image gallery showing unspecific staining of cells analyzed with the DEPArray™. ... 74
Figure 4.3. Flow cytometry analysis of WBCs isolated from HD blood samples collected in EDTA or in Streck tubes and stained with distinct antibodies directed against EpCAM. ... 76
Figure 4.4. Image gallery showing staining and morphological characteristics of cells analyzed with the DEPArray™. ... 78
Figure 4.5 DEPArray™ image gallery showing MCF7 cells identified in a spiked-in HD blood sample. ... 80
List of figures and tables Figure 4.9. CTCs, WBCs and double-negative cells identified in two blood samples from
biliary tract cancer patients. ... 89
Figure 4.10. Copy number alteration profiles of control DNA samples. ... 92
Figure 4.11. Number of patients enrolled and included in each type of analysis. ... 94
Figure 4.12. Identification of eCTCs, ncCTCs and WBCs in clinical samples. ... 95
Figure 4.13. Association between eCTCs and survival. ... 97
Figure 4.14. Effect of treatment on CTCs in BTC patients... 99
Figure 4.15. CTC monitoring during treatment. ... 101
Figure 4.16. Vimentin expression in eCTCs. ... 102
Figure 4.17. MCF7 cells identified in a spiking experiment. ... 103
Figure 4.18. Non-aberrant, circulating double-negative cells expressing CSV. ... 104
Figure 4.19. Mutational profiling of 9 CTCs recovered from one blood sample from a BTC patient. ... 106
Figure 4.20. CTCs’ chromosomal instability in responding and non-responding patients. ... 110
Figure 4.21. Clustering analysis of CNA profiles of single CTCs isolated from BTC patients. ... 111
Figure 4.22. Comparison of CNA frequency in 2 specific chromosome regions, detected in CTCs from clusters 2 and 3. ... 112
Figure 4.23. Detection of HuCCT-1 cells in HD WBCs by RNA analysis. ... 116
Figure 4.24. Detection of HuCCT-1 cells in HD WBCs by RNA analysis after transcripts’ pre-amplification. ... 117
Figure 4.25. DPcells detected in BTC patients. ... 118
Figure 4.26. DPcells and DP-CTCs distribution in clinical samples. ... 119
Figure 4.27. Combined DPcell- and CTC- stratification in association with progression-free survival. ... 121
Table 3.1. Antibodies used to stain surface markers. ... 62
Table 3.2. Antibodies used to stain intracellular markers. ... 63
Table 4.5. Antibodies added to the protocol and tested for sensitivity and specificity. ... 86 Table 4.6. Patients’ clinico-pathological characteristics. ... 93 Table 4.7. Number of eCTCs and ncCTCs detected for each analyzed blood sample... 96 Table 4.8. Association between eCTC-status and clinical characteristics of the patients
included in the survival analysis. ... 98 Table 4.9. Variations in CTC numbers during treatment. ... 100 Table 4.10. LST scores of CTCs recovered from patients responding to treatment. ... 108 Table 4.11. LST scores of CTCs recovered from patients non-responding to treatment. . 109
Abbreviations
ADO Allele drop out
ALK Anaplastic lymphoma kinase AR Androgen receptor
BL Baseline
BTC Biliary tract cancer CCA Cholangiocarcinoma CI Confidence interval cis/gem Cisplatin plus gemcitabine CK Cytokeratins
CNA Copy number alteration CSV Cell surface vimentin CTC Circulating tumor cell DPcell Dual-positive cell DSS Disease-specific survival
DT During treatment eCTC Epithelial CTC
EGFR Epithelial growth factor receptor EMT Epithelial to mesenchymal transition EOT End of treatment
EpCAM Epithelial cell adhesion molecule EphA3 Ephrin type-A receptor 3
FBS Fetal bovine serum
FGFR Fibroblast growth factor receptor
FOLFOX Oxaliplatin, L-folinic acid and 5-fluorouracil FU Follow-up
GBC Gallbladder cancer GII Genome integrity index HD Healthy donor
HER2 Human epidermal growth factor receptor 2 HR Hazard ratio
IDH1 Isocitrate dehydrogenase 1 INDELs Small insertions and deletions KRT8 Cytokeratin-8
LM-PCR Ligation-mediated PCR
lp-WGS Low-pass whole-genome sequencing LST Large-scale state transition
MALBAC Multiple annealing and looping-based amplification MDA Multiple displacement amplification
ncCTC Non-conventional CTC NGS Next generation sequencing OS Overall survival
Abbreviations
PR Partial response QC Quality control RBC Red blood cell SD Stable disease
SNP Single nucleotide polymorphism VEGF Vascular endothelial growth factor VIM Vimentin
WBC White blood cell
Introduction
1.1.
Biliary tract cancer
Biliary tract cancer (BTC) consists of different epithelial malignancies arising in any part of the biliary tree and includes cholangiocarcinoma (CCA), gallbladder cancer (GBC) and ampulla of Vater cancer. According to the location, cholangiocarcinomas are subdivided into intrahepatic and extrahepatic CCA (located in the intrahepatic and extrahepatic bile ducts, respectively) and the latter can be further divided into distal and hilar CCA (Figure 1.1) [Tariq N.U. et al., 2019].
Figure 1.1. Anatomical sub-variants of BTC.
According to the location of the tumor, BTCs are subdivided into gallbladder cancer (GBC), ampulla of Vater cancer (AVC), intrahepatic cholangiocarcinoma (IHC) and extrahepatic cholangiocarcinoma (EHC), further subdivided into perihilar and distal extrahepatic cholangiocarcinoma. [Tariq N. et al.,2019]
rare disease (< 6 cases per 100,000 people) [Banales J.M. et al., 2016], its incidence is exceptionally high in some Eastern countries (up to 85 cases per 100,000 people for northeast Thailand) due to different geographical risk factors and genetic determinants [Khan S.A. et al., 2019]. In 2017, the global BTC incidence was 211,000 cases, with 174,000 deaths [Global Burden of Disease Collaboration, 2019] and, over the past decades, both its incidence and its mortality have increased worldwide, in particular with regards to intrahepatic CCA [Taylor-Robinson S.D. et al., 2001; Patel T., 2002; Khan S.A. et al., 2002; Bertuccio P. et al., 2013; Global Burden of Disease Cancer Collaboration, 2015; Bertuccio P. et al., 2019].
BTCs are aggressive diseases characterized by a poor prognosis (5-years survival rate = 5-15%, considering all stages) [Anderson C. and Kim R., 2009; Lamarca A. et al., 2020]. Moreover, since they are generally asymptomatic in early stages, most BTCs are diagnosed at metastatic stage, when the 5-year survival rate is only 2% [Tariq N.U. et al., 2019].
Introduction
1.1.1.
Clinical management of BTC: State of the art
Currently, the treatment of BTC is not based on the anatomical subtypes, but solely on the stage of the disease and it essentially consists of surgery and systemic chemotherapy (Figure 1.2) [Valle J.W. et al., 2016].
Figure 1.2. Algorithm for the management of patients with BTC.
The ESMO guidelines for the treatment of early-stage, locally-advanced and metastatic biliary tract cancer are reported. MDT, multidisciplinary team; PS, performance status; iCCA, intrahepatic cholangiocarcinoma. [Valle J.W. et al., 2016]
Unfortunately, though, only approximately 20% of patients present an early stage disease at diagnosis and are eligible for surgery [Bridgewater J. et al., 2014]. Moreover, the majority of patients undergoing surgical resection will relapse, predominantly developing liver metastasis [Miyazaki M. et al., 2017]. Until recently, the benefit derived from the use of adjuvant therapies in reducing BTC relapse and improving patient survival was not clear due to the lack of randomized clinical trials dedicated to BTC [Lamarca A. et al., 2020]. In 2019, the BILCAP trial, a randomized phase III trial comparing the use of adjuvant capecitabine to observation alone in BTC patients who had undergone a complete resection with curative intent, was completed and showed an improvement in median overall survival (OS) for patients with BTC receiving capecitabine as adjuvant chemotherapy (36 to 53 months for patients in the observation group vs. capecitabine group, respectively, p = 0.028) [Primrose J.N. et al., 2019]. Although the relapse rate remained high in patients treated with capecitabine (60%), the results of the BILCAP study have led to a change in the international guidelines for patients with resected BTC, which now recommend adjuvant capecitabine for 6 months after surgery as standard of care [Shroff R.T. et al., 2019].
However, for the vast majority of BTC patients (including those experiencing recurrence after surgery and those diagnosed with locally-advanced or metastatic disease), palliative systemic treatments represent the only option. Clinical trials have shown that systemic chemotherapy, compared with best supportive care, extends patients’ survival [Glimelius B. et al., 1996, Sharma A. et al. 2010] and, in 2010, cisplatin combined with gemcitabine (cis/gem) became the first-line standard of care treatment for patients with advanced BTC [Banales J.M. et al., 2016], supported by the results of two randomized trials conducted in the United Kingdom (ABC-02, phase III) and in Japan (BT22, phase II). Both the ABC-02
Introduction superiority of the combination over gemcitabine alone, with an increased median OS for the patients treated with cis/gem of 11.7 vs. 8.1 months (hazard ratio (HR) = 0.64; 95% confidence interval (CI): 0.52–0.80) and 11.1 vs. 7.7 months (HR = 0.69; 95% CI: 0.42– 1.13) for the ABC-02 and the BT22 study, respectively. These results were further confirmed by a combined meta-analysis of the 2 trials [Valle J.W. et al., 2014] showing a significant improvement in progression-free survival, PFS, (HR = 0.64, 95% CI 0.53–0.76,
p < 0.001) and OS (HR = 0.65, 95% CI: 0.54–0.78, p < 0.001) for cis/gem over gemcitabine alone, independent of patient age and gender, primary tumor site, prior therapy, stage of disease and ethnicity. The meta-analysis also revealed that patients with poor performance status (performance status = 2) derived the least benefit from cis/gem, suggesting that gemcitabine alone could be the best option for these patients.
Nonetheless, the effectiveness of cis/gem treatment is limited (OS < 1 year) and most patients will develop resistance and will undergo a second-line therapy. Unfortunately today there is no established second-line regimen available. In fact, different drugs have been tested over the years to treat patients with BTC progressing to first-line chemotherapy, but a systematic review [Lamarca A. et al., 2014] of 25 studies on a total of 761 patients receiving second-line systemic chemotherapy (including 14 phase II clinical trials, 9 retrospective analyses and 2 case reports) reported disappointing median PFS (3.2 months; 95% CI: 2.7-3.7) and response rate (7.7%; 95% CI: 4.6-10.9) and did not find sufficient evidence to recommend any second-line treatment. However, an ongoing randomized phase III trial (ABC-06) is comparing active symptom control alone or in combination with oxaliplatin, L-folinic acid and 5-fluorouracil (FOLFOX) as second-line therapy for advanced BTC patients. The initial results of the trial, presented last year at the ASCO conference [Lamarca A. et al., 2019], showed an increased survival after 12 months of treatment, going from 10% to 25%, for patients receiving active symptom control alone
second-line therapy is currently recommended, FOLFOX is frequently used as second-line treatment in BTC patients.
1.1.2.
Clinical management of BTC: Future challenges
Considering the modest therapeutic efficacy of chemotherapy in BTC, new therapies are urgently needed.
The first attempts to improve patient outcome by using targeted agents were focused on the vascular endothelial growth factor (VEGF) and on the epithelial growth factor receptor (EGFR) axes. In fact, both angiogenesis and the overexpression of the EGFR family have been implicated in BTC carcinogenesis [Wehbe H. et al., 2006; Pignochino Y. et al., 2010; Simone V. et al., 2017; Adeva J. et al., 2019]. Different phase II trials have tried to assess the benefit of targeting the VEGF pathway using antibodies (bevacizumab) [Zhu A.X. et al., 2010; Iyer R.V. et al., 2018; Larsen F.O. et al., 2018] or tyrosine kinase inhibitors (sorafenib, cediranib and vandetanib) [Moehler M. et al., 2014; Valle J.W. et al., 2015; Santoro A. et al., 2015] either alone or in combination with chemotherapy, but failed to produce evidence supporting the use of anti-angiogenic agents. Similarly, trials using EGFR inhibitor [Lee J. et al., 2012] or anti-EGFR antibody [Sohal D.P. et al., 2013; Malka D. et al., 2014; Chen J.S. et al., 2015] in combination with chemotherapy did not produce encouraging results. The lack of positive results, though, could be due to the fact that these trials were conducted in unselected populations.
Actually, even though in the past BTCs have been therapeutically managed as a single disease focusing only on the disease stage, recent technological advances (such as next generation sequencing, NGS) have revealed a remarkable molecular complexity of BTC, showing not only the presence of numerous alterations in putative driver and actionable
Introduction anatomical subtype [Nakamura H. et al., 2015; Jusakul A. et al., 2017]. In particular, alterations in IDH1/2, EPHA1, BAP1 and FGFR2 were more frequently found in intrahepatic CCA, whereas gene fusions involving PRKACA or PRKACB and genetic aberration in ARID1A, PI3KCA and the ERBB family were detected in extrahepatic CCA (Figure 1.3); GBC was instead characterized by ERBB3 and EGFR mutations [Braconi C. et al., 2019].
Figure 1.3. Molecular spectrum of intrahepatic and extrahepatic CCA.
The most unique and prevalent genetic alterations found in different anatomical locations are reported. iCCA, intrahepatic cholangiocarcinoma; pCCA, perihilar cholangiocarcinoma; dCCA, distal cholangiocarcinoma; eCCA, extrahepatic cholangiocarcinoma. [Braconi C. et al., 2019]
These results highlighted the high heterogeneity of BTC and the need of including molecular profiling for clinical decisions. Through BTC sequencing analyses, promising new molecular targets have been identified and new trials have been initiated to explore their potential in BTC [Bogenberger J.M. et al., 2018; Rizvi S. et al., 2018]. In particular
FGFR fusions, respectively), are ongoing, while other targeted agents (such as anti-HER2 and PARP inhibitors) are in different stages of clinical development [Adeva J. et al., 2019]. However, as the information about BTC genomic aberrations expanded, it also revealed partially conflicting data, possibly due to differences in study populations, use of different detection technologies, and misclassifications [Kendall T. et al., 2019]. Further studies are needed to accomplish more accurate molecular profiling data, and to reach a better understanding of the differences between anatomical subtypes and between subgroups within the subtypes. Moreover, in a recent study on 4 patients with intrahepatic CCA [Walter D. et al., 2017], different samples from the same tumor were analyzed for the detection of private and common mutations to evaluate intra-tumor heterogeneity. Private mutations were identified in 3 out of 4 patients and, across all samples, the mean percentage of private mutations per sample was 12%, indicating a high level of intra-tumor heterogeneity, limiting the potential applicability of personalized medicine in BTC. On the other hand, high intra-tumor heterogeneity could be indicative of sensitivity to immune checkpoint inhibitors [McGranahan N. et al., 2016]. In fact, immunotherapy is a promising therapeutic strategy in BTC and it is currently under evaluation in different phase I and II trials [Tariq N.U. et al., 2019; Kelley R.K. et al., 2020]. Intra-tumor heterogeneity in intrahepatic CCA was investigated also in another study [Dong L-Q. et al., 2018] in which multi-regional whole-exome sequencing was performed for 6 patients and showed the presence, in all patients’ tumors, of different subclones, supporting a parallel evolution model. Moreover, for one patient from whom samples of both the primary tumor and the recurrence were available, the authors observed a subclonal structure also in the recurrence samples and the presence of private mutations at recurrence, suggesting a polyclonal metastatic seeding and clonal evolution (Figure 1.4).
Introduction
Figure 1.4. Polyclonal seeding and tumor evolution in BTC.
(A) Timeline of the clinical history of patient ICC1239 from whom tissue samples were available at the time of primary resection (1239P) and the time of intrahepatic recurrence (1230R). CT/MRI images of each tumor are reported; the tumors are indicated by the red arrows. (B) Phylogenetic tree showing alterations shared by all biopsies (green), private alterations of the recurrence (orange) and private alterations of the primary tumor (grey). [Adapted from Dong L-Q. et al., 2018]
These data suggest that performing a single biopsy of the primary tumor for therapeutic decisions could not be enough and could be an explanation for the limited efficacy of targeted therapies in BTC, even in highly selected populations: In the BGJ398 phase II study, a FGFR kinase inhibitor was given to selected intrahepatic CCA patients with FGFR alterations or FGFR2 fusions but an overall response rate of only 15% was reached [Javle M. et al., 2018].
Another aspect that has to be taken into consideration is that most sequencing data produced so far were obtained by the analysis of resected tumors, while the majority of patients present with an inoperable tumor at diagnosis. Considering that tumors can evolve
early stage tumors reflect the genomic landscape of advanced tumors (Figure 1.5) [Braconi C. et al., 2019].
Figure 1.5. Changes occurring in the molecular landscape of advanced BTC.
Polyclonal metastatic seeding, emergence of new clones in response to treatment, and evolution of the tumor over time (causing the loss of subclones present in the primary tumor) are mechanisms potentially causing differences in the molecular portraits of primary vs. advanced BTCs. [Braconi C. et al., 2019]
There is, therefore, the need to deepen our knowledge on the molecular characteristics of advanced BTC. Unfortunately, patients with advanced BTC are generally diagnosed by cytology or small biopsies and it is often impossible to obtain enough tumor material to perform genomic characterization analyses. One possibility to overcome the lack of tumor tissue from these patients could be the use of liquid biopsies. In particular, circulating tumor cells (CTCs) could function as an alternative source of tumor material, giving new hints on molecular drivers in advanced BTC for the development of new therapeutic strategies.
CTCs could also be used to study tumor evolution and the development of therapeutic resistance over time, without the need of performing repeated tissue biopsies (an invasive procedure which is even more difficult to perform in BTC patients, due to the anatomical location of the tumor). In fact, BTCs are characterized by a remarkable resistance to both
Introduction mechanisms of chemoresistance that confer to BTC cells the so-called multi-drug resistance phenotype [Marin J.J.G. et al., 2018]. Given the high heterogeneity of BTCs, the genetic signatures underlying the mechanisms of chemoresistance are also diverse and can vary over time, in response to pharmacological treatment, contributing to the acquired chemoresistance [Fouassier L. et al., 2019]. Understanding the molecular bases of chemoresistance and how these evolve during treatment would allow for the prediction of the potential failure of a given treatment and would support the choice of the best therapy for each patient at a specific time. This can only be attained by getting multiple tumor biopsies over time from the same patient, which is hardly feasible in BTC patients. In this context, the analysis of CTCs could really constitute a pivotal point in the advance of our knowledge on BTC’s molecular changes during disease progression.
1.2.
Circulating tumor cells
In patients with solid tumors, CTCs are released from both the primary tumor and the metastatic lesions into the bloodstream during the course of the disease. Different technologies allow the detection and the characterization of CTCs, which are therefore considered a real-time liquid biopsy of tumors [Pantel K. and Alix-Panabieres C. 2010]. The term liquid biopsy includes also the analysis of other tumor-derived elements circulating in the blood, such as circulating tumor DNA (ctDNA), tumor-derived exosomes and microvesicles, tumor-educated platelets, circulating tumor micro RNA, mRNA and non-coding RNA [Junqueira-Neto S. et al., 2019; Keller L. and Pantel K., 2019], each of which can provide different and complementary information. CTCs, in particular, being intact and viable cells, offer the possibility of performing a multilevel analysis of genotype (DNA) and phenotype (RNA and proteins). Moreover, they are a highly selected subpopulation of tumor cells, able to leave the primary tumor and survive in the bloodstream (the majority of CTCs die soon after entering the blood vessels, due to anoikis, attack by cells of the immune system and fluid shear stress [Berezovskaya O. et al., 2005; Huang Q. et al., 2018]), suggesting that CTCs could be representative of the most aggressive clones of the tumor [Pantel K. and Alix-Panabieres C. 2013]. Overall, CTC analysis can potentially be used for i) early detection of cancer, ii) prognostic stratification of patients, iii) identification of therapeutic targets, iv) prediction of response to targeted treatments, v) treatment monitoring and vi) identification of resistance mechanisms (Figure 1.6) [Alix-Panabieres C. and Pantel K. 2016]. Whereas for the latter applications a large amount of data have been produced (and will be discussed subsequently), data regarding the use of CTCs for early detection are scanty and limited to lung cancer. In a study by Ilie and colleagues, CTCs could be detected in a small
Introduction 2014]. Nonetheless, in a following study by the same investigators [Marquette C.H. et al., 2020], the detection of CTCs in high-risk patients was unable to predict lung cancer development. The applicability of CTCs for the early detection of cancer is therefore still an open question.
and to monitor the efficacy of treatment and the development of resistance in real-time. ER+, estrogen receptor positive. [Alix-Panabieres C. and Pantel K., 2013]
However, the study of CTCs is extremely technically challenging because they are rare events diluted in billions of blood cells. In order to detect and analyze CTCs it is therefore necessary to use technologies to enrich and identify them.
1.2.1.
First generation of CTC studies: CTC enumeration
The first generation of CTC studies used methodologies for CTC detection based on the expression of epithelial markers (such as epithelial cell adhesion molecule, EpCAM and cytokeratins, CK). These markers are in fact expressed by epithelial tumors but not by white blood cells (WBCs) and allow the discrimination between CTCs, i.e. cells positive for EpCAM/CK and negative for the pan-leukocyte marker CD45, and WBCs, i.e.
EpCAM/CK-negative and CD45-positive cells [Alix-Panabieres C. and Pantel K., 2013]. The gold-standard instrument for CTC detection based on EpCAM and CK expression is the CellSearch® system [Allard W.J. et al., 2004], which, starting from 7.5 mL of blood, performs an enrichment step of CTCs using ferrofluids conjugated with antibodies against EpCAM, followed by an identification step in which the expression of CK and CD45 in the enriched cells is evaluated by staining them with fluorescently-labeled antibodies, to allow for CTC identification and counting. By using the CellSearch® system, in 2004 Cristofanilli and colleagues [Cristofanilli M. et al., 2004] demonstrated, for the first time, the prognostic significance of CTCs in metastatic breast cancer patients: Patients with ≥ 5 CTCs per 7.5 mL of blood had a shorter median PFS (2.7 months vs. 7.0 months,
p < 0.001) and OS (10.1 months vs. >18 months, p < 0.001), than patients presenting < 5 CTCs. Successively, the prognostic relevance of CTC enumeration by CellSearch® was further confirmed by a pooled analysis of 20 studies including data from around 2000
Introduction al., 2014; Bidard F.-C. et al., 2018], colorectal cancer [Huang X. et al., 2015; van Dalum G. et al., 2015; Bork U. et al., 2015], metastatic prostate cancer [de Bono J.S. et al., 2008], small-cell [Hou J.M., et al., 2012; Tay R.Y., et al., 2019] and non-small-cell [Krebs M.G., et al., 2011; Lindsay C.R., et al., 2019] lung cancer. Moreover, changes in CTC counts during therapy have been shown to have a prognostic significance in patients with metastatic breast [Hayes D.F. et al., 2006], colorectal [Cohen S.J., et al., 2008], prostate [Lorente D. et al., 2016] and lung [Hou J.M., et al., 2012] cancer, thus supporting the use of CTC enumeration for treatment monitoring as well (Figure 1.7) [Cabel L. et al., 2017].
Figure 1.7. Clinical validity of CTC enumeration by CellSearch®.
CTC counts have been shown to be associated with prognosis in patients with breast, colorectal and metastatic prostate cancer. The detection of variations in CTC numbers during therapy can be used for treatment monitoring. [Adapted from Cabel L. et al., 2017]
This evidence led to the approval by the Food and Drug Administration of the CTC enumeration by CellSearch® to predict prognosis and monitor treatment in patients with metastatic breast, colorectal and prostate cancer. In 2010, CTCs were included in the
lymphnodes [Alix-Panabieres C. and Pantel K., 2014], and in 2019 an expert consensus paper [Cristofanilli M. et al., 2019], performing a pooled analysis of almost 2500 patient data, recommended CTC enumeration as a new tool to improve prognostic stratification of metastatic breast cancer in Stage IVindolent and Stage IVaggressive (for patients presenting < 5
CTCs and ≥ 5 CTCs, respectively).
The clinical utility of CTC enumeration was first assessed by the SWOG S0500 study [Smerage J.B. et al., 2014], a trial designed to test whether, in metastatic breast cancer patients undergoing first-line chemotherapy, persistently high levels of CTCs after the first cycle of therapy could be used to identify the patients who would benefit from switching to a new chemotherapy regimen. The study reported no survival improvement for patients who switched therapy, suggesting that CTCs were not helpful in driving therapeutic intervention. However, the negative results of the trial were probably due to a design flaw [Rossi E. and Fabbri F., 2019]. In fact, in metastatic breast cancer, patients not responding to first-line therapy are often markedly resistant to subsequent lines of therapy [Bidard F.-C. and Pierga J.-Y., 2015] and changes in first-line chemotherapy introduced on the base of imaging evaluation did not improve patients’ OS as well [Alunni-Fabbroni M. et al., 2014]. Therefore, the clinical utility of CTCs is still an open question and there are several clinical trials currently evaluating if treatment decisions can be improved by considering either CTC count (CirCe01 and STIC-CTC studies) or CTC phenotype (DETECT III and DETECT IV trials) [Schocter F. et al., 2019]. While most studies are still ongoing, the results of the STIC-CTC trial were presented at the San Antonio Breast Cancer Symposium in 2018 and showed that, in patients with human epidermal growth factor receptor 2 (HER2)-negative hormone receptor-positive metastatic breast cancer, CTC-driven administration of chemotherapy as first-line treatment resulted in a
Introduction needed, these promising results support the clinical utility of CTCs, at least in selected cohorts of cancer patients.
1.2.1.1.
CTC enumeration in patients with BTC
The clinical relevance of CTCs in patients with BTC was investigated in a few studies by using the CellSearch® [Al Utswani O. et al., 2012; Valle J.W. et al., 2015; Yang J.D. et al., 2016; Backen A.C. et al., 2018]. As observed for other malignancies, the presence of CTCs was associated with poor prognosis also in BTC [Valle J.W. et al., 2015; Yang J.D. et al., 2016]. In the ABC-03 study [Valle J.W. et al., 2015], this result was confirmed by using both 1 cell or 2 cells per sample as positivity threshold, showing a significantly higher risk of death for patients presenting 1 or ≥ 2 CTCs, than patients with 0 CTCs (1 CTC vs. 0 CTC, HR = 3.25; 95% CI: 1.81–5.83; ≥ 2 CTCs vs. 0 CTC, HR = 3.00; 95% CI: 1.73–5.22,
p < 0.0001), indicating that the presence of even 1 CTC is clinically relevant in this clinical setting (Figure 1.8).
However, in the ABC-03 trial, CTCs, although associated with prognosis, were not treatment-predictive [Backen A.C. et al., 2018]. Moreover, the reported CTC detection rates in all the studies were quite low, ranging from 17% - 25% [Al Utswani O. et al., 2012; Yang J.D. et al., 2016] to 47% [Valle J.W. et al., 2015] by using 2 CTCs and 1 CTC as positivity threshold, respectively. These results suggest that in the blood of BTC patients, as already reported for other epithelial cancers, there could be a subpopulation of CTCs which is not detected by CellSearch®, due to a non-epithelial phenotype [Backen A.C. et al., 2018]. Further studies using innovative technologies also capable of detecting these CTCs are therefore needed in order to understand the role of CTCs in BTC.
1.2.2.
Technologies to capture CTC heterogeneity
Even though the first strategy used to detect CTCs (based on the expression of epithelial markers) led to the assessment of the clinical validity of CTCs in some cancers (including breast, prostate, colorectal and lung cancer), it was not equally effective in other tumors of epithelial origin (such as ovarian, kidney and biliary tract cancer) or in demonstrating CTCs’ clinical utility [Raimondi C. et al., 2014]. This could be due to the presence of CTCs with a non-epithelial phenotype, as suggested by the fact that CellSearch® detects CTCs only in a proportion of advanced cancer patients, and by the reported presence of CK-negative/CD45-negative cells in patients’ blood [Mego M. et al., 2011; Gazzaniga P. et al., 2011; Wang L. et al., 2016]. CTCs can, in fact, lose the expression of epithelial markers through the epithelial to mesenchymal transition (EMT), a process involved in carcinoma invasion and metastasis [Paterlini-Brechot P. and Benali N.L., 2006; Lee J.M. et al., 2006]. EMT is not an “all or nothing” process, but rather induces a variety of
Introduction with a mesenchymal [Zhang L. et al., 2013; Satelli A. et al., 2015b; Xu L. et al., 2017] or a mixed epithelial-mesenchymal [Lecharpentier A. et al., 2011; Giordano A. et al., 2012; Yokobori T. et al., 2013; Morrow C.J. et al., 2016; Bulfoni M. et al., 2016] phenotype have in fact been reported in cancer patients, indicating a much higher phenotypic heterogeneity in CTCs than originally thought. Moreover, it has been demonstrated that CTCs’ phenotype can change over the course of the disease, and these changes can be related to treatment resistance [Yu M. et al., 2013; Tsao S.C. et al., 2018]. In particular, Yu and colleagues, by evaluating the expression of epithelial and mesenchymal markers in CTCs from 10 patients with metastatic breast cancer undergoing treatment, reported an increase in CTCs with mesenchymal features after treatment in patients undergoing progression, whereas, in patients responding to treatment, a decrease in the CTC number and/or in the proportion of mesenchymal CTCs was observed, suggesting an association between mesenchymal CTCs with disease progression (Figure 1.9).
(M>E; M=E; E>M). For each patient, the fractionation of CTCs, according to E/M ratios, is reported for both time points. The breast cancer subtype, the treatment regimen (chemotherapy or targeted therapies) and the number of days under treatment are also reported. [Adapted from Yu M. et al., 2013]
Based on the ascertained phenotypic heterogeneity of CTCs, over the last several years a wide variety of new technologies have been developed to try to detect all CTC subpopulations [Millner L.M. et al., 2013; Ferreira M.M. et al., 2016; Batth I.S. et al., 2019]. All the different methodologies exploited 3 main strategies: i) to expand the range of markers used for positive selection, by additionally including mesenchymal markers as the cell surface vimentin, CSV, [Satelli et al., 2015a], stem cell markers as CD133 [Guo W. et al., 2018], and cancer specific antigens as HER2 and the prostate-specific membrane antigen, PSMA [Galletti G. et al., 2014; Kirby B.J. et al., 2012]; ii) to perform a negative depletion of blood cells using antibodies against CD45 [Bluemke K. et al., 2009; Giordano A. et al., 2012]; iii) to use physical properties for distinguishing between CTCs and WBCs such as size, density and deformability, through filtration [Desitter I. et al., 2011; Ma Y.-C. et al., 2013], density gradient centrifugation [Muller V. et al., 2005; Hodgkinson C.L. et al., 2014], or the use of microfluidic devices [Che J. et al., 2016; Miller M.C. et al., 2018]. New CTC detection technologies also moved forward from the simple enumeration and focused on the characterization of detected cells as well, to allow both a deeper understanding of the role of the different CTC subpopulations and to increase the sensitivity of the detection methods (Figure 1.10). The abovementioned detection strategies have, in fact, an increased sensitivity but at the same time can detect a high number of false positive events (such as normal cells expressing positive selection markers; circulating CD45-negative cells that are not CTCs, as endothelial cells; and WBCs with physical characteristic similar to CTCs). Due to the specificity issue of these methods,
Introduction
Figure 1.10. New methods for CTC detection.
The new methods for CTC enrichment are based on physical properties, positive selection or negative selection. Generally, the methods include also a characterization step analyzing CTCs’ DNA, RNA or proteins, as well as their functional properties. [Cabel L. et al., 2017]
With regards to characterization approaches, recent technological advances (such as instruments for single-cell isolation and NGS) have allowed the development of methods for the genomic analysis of CTCs at the single-cell level (Figure 1.11) [Lim S.B. et al., 2019]. These methods include, after the enrichment, a step for the isolation of single cells. Cells can be individually isolated by laser capture microdissection [Park E.S. et al., 2018] and fluorescence-activated cell sorting (FACS) [Lambros M.B. et al., 2018], or by using specific instruments such as the DEPArray™ which isolates single cells by exploiting dielectrophoresis [Peeters D.J. et al., 2013], microscopic manipulators as the CellCelector™ [Lampignano R. et al., 2017; Reinhardt F. et al., 2019] and microfluidic devices [Yeo T. et al., 2016; Valihrach L. et al., 2018].
Figure 1.11. Standard workflow for single-CTC sequencing.
After enrichment from whole blood, CTCs are individually isolated and undergo WGA. The quality of WGA products is checked, and high quality amplified DNA can be sequenced for the detection of small-scale and large-scale alterations. Quality checks are performed on the sequencing library to assess the possibility of amplification bias. LCM, laser capture microdissection; FACS, fluorescence-activated cell sorting; DOP-PCR, degenerate oligonucleotide primed PCR; LA-DOP-PCR, linker-adapter PCR; MDA, multiple displacement amplification; MALBAC, multiple annealing and looping-based amplification cycles; SNVs, single nucleotide variants; MSI, microsatellite instability; CNVs, copy number variations, LSTs, large-scale state transitions; ADO, allelic drop out; QC, quality control. [Lim S.B. et al., 2019]
Independently from the type of isolation method used, all isolated cells will undergo the whole-genome amplification (WGA) in order to be analyzed. WGA is based on 2 approaches: PCR and multiple displacement amplification (MDA), and the different WGA methods employ either only one of the 2 (PCR-based and MDA-based) or a combination of both mechanisms (hybrid) [Blainey P.C., 2013]. PCR-based methods always require a fragmentation of the genome before the amplification, due to the limited processivity of the polymerase (generally a Taq DNA polymerase), and use, for the amplification step, either random primers that anneal across the genome in a non-deterministic manner (degenerate oligonucleotide primed PCR, DOP-PCR), or specific sequence primers that are ligated at the end of the template DNA fragments (linker adapter, or ligation-mediated PCR,
Introduction thanks to the use of a restriction endonuclease, allowing a defined and reproducible fragmentation and priming pattern and a high genomic coverage [Czyz Z.T. and Klein C.A., 2015]. MDA-based methods instead use highly processive DNA polymerases (such as the Phi29) which perform an isothermal rolling circle amplification of the template DNA, yielding a high quantity of DNA (20-30 µg per single cell) and producing large amplicons of > 10 kb [Dean F.B. et al., 2002]. Hybrid approaches, as the PicoPLEX® [Kamberov E. et al., 2004] and the multiple annealing and looping-based amplification (MALBAC) [Zong C. et al., 2012], combine an initial MDA-based pre-amplification step, followed by the amplification of the produced amplicons by PCR. All the WGA approaches, unfortunately, have biases which can result in a non-homogeneous amplification of all amplicons (due to random events occurring during the initial steps of the amplification, or to the intrinsic characteristics of the amplicons, such as the GC content) and in the production of artifacts related to the polymerase activity [Sabina J. and Leamon J.H., 2015]. These factors can result in allele drop out (ADO), loss of coverage, low coverage uniformity, or allelic imbalances. Each method is characterized by advantages and limitations, as reported by studies which have compared different commercially available WGA kits based on LM-PCR (Ampli1™), MDA (REPLI-g®), or a
combination of both (PicoPLEX® and MALBAC), sometimes obtaining contrasting results [Czyz Z.T. et al., 2015; Babayan A. et al., 2016; Borgstrom E. et al., 2017; Deleye L. et al., 2017]. Overall, REPLI-g® (MDA-based WGA) showed the least sensitivity and coverage uniformity and the highest ADO rate [Czyz Z.T. et al., 2015; Babayan A. et al., 2016; Borgstrom E. et al., 2017], although in the study by Deleye and colleagues it was reported as an efficient method, comparable to the other approaches [Deleye L. et al., 2017]. LM-PCR-based (Ampli1™) and hybrid (PicoPLEX® and MALBAC) methods have
suited for the detection of single nucleotide polymorphisms (SNPs) and small insertions-deletions (INDELs) [Babayan A. et al., 2016], whereas hybrid approaches seemed to be the best choice for copy number alteration (CNA) analysis [Babayan A. et al., 2016]. However, the applicability of Ampli1™ for CNA analysis was demonstrated by studies
showing the possibility of detecting known CNAs in single cells undergoing Ampli1™
WGA [Klein C.A. et al., 1999; Schmidt-Kittler O. et al., 2003; Mohlendick B. et al., 2013]. WGA is not always successful due to the possibility of losing cells during the isolation procedure and to the quality of the isolated cells' DNA. Therefore, after WGA, quality control (QC) assays are performed to assess the DNA yield and the length of amplified fragments. In particular, Polzer and colleagues [Polzer B. et al., 2014] developed a QC assay based on a multiplex-PCR approach which assigns a score called genome integrity index (GII) to WGA products. The GII is predictive of successful application of different downstream sequencing analyses: Good quality samples can undergo any type of sequencing analysis (including Sanger sequencing, array comparative genomic hybridization (aCGH) platforms or genome-wide NGS) for the detection of a variety of genomic alterations (Figure 1.12) including small-scale alterations (single nucleotide variants (SNVs), INDELs and microsatellite instability), and large-scale alterations (copy number variations (CNVs), chromosomal breakpoints or large-scale state transitions (LSTs), and chromosomal rearrangements).
Introduction
Figure 1.12. Genomic alterations in single CTCs.
List and characteristics of the type of genomic alterations that can be detected in CTCs by single-cell sequencing. [Adapted from Lim S.B. et al., 2019]
The methods for CTC characterization at the single-cell level opened a new chapter of liquid biopsy research, aimed at characterizing and monitoring changes in tumor heterogeneity in individual patients to further understand the biology of tumor evolution.
1.2.3.
New generation of CTC studies: CTC characterization at the
single-cell level
The studies focused on the analysis of CTCs at the single-cell level have tried to answer 3 main questions:
Do CTCs recapitulate tumor spatial heterogeneity?
Can we use CTCs to study the evolution of tumor heterogeneity during the course of the disease?
Can CTC molecular characterization provide information on treatment resistance? In 2014 Lohr and colleagues conducted the first proof-of-concept study showing the possibility to recapitulate tumor heterogeneity by analyzing CTCs [Lohr J.G. et al., 2014]. The authors performed whole-exome sequencing of 19 single CTCs, 1 metastatic biopsy and 9 spatially different primary tumor biopsies from a single patient with metastatic prostate cancer. Fifty-six metastatic trunk mutations (mutations present in both the metastasis and the primary tumor) were identified and 73% of these were detected in CTCs as well. CTCs also carried 51% of all the mutations detected in the metastatic site. Moreover, by hierarchical clustering it was possible to assess the similarities between the samples collected from different anatomical locations (primary tumor, lymph node and blood) over the entire course of the disease (from the time of the radical prostatectomy to the blood collection, 5.3 years later), allowing to reconstruct an evolutionary tree of the tumor (Figure 1.13). The 9 primary tumor biopsies showed high intra-tumor heterogeneity, but there was one region that closely resembled CTCs and the metastatic biopsy, suggesting that this particular region of the primary tumor could be the one responsible for the metastatic spread.
Introduction
Figure 1.13. Dendrogram representing hierarchical clustering and the timeline of sample acquisition.
Somatic mutations in the listed genes were detected in all individual biopsies of tissue (early trunk) or in all biopsies belonging to only one of the two branchpoints (the nonsynonimus mutations are indicated in bold, with asterisks). The pathology blocks from which tissue biopsies were obtained are represented in pink (drawn to scale) and, within each block, the area with histological presence of tumor is shown by the dotted line. The colors highlight the site where each individual biopsy of the dendrogram was taken. Below, a schematic representation of the regions of the prostate from which the pathology blocks were obtained, is reported. [Adapted from Lohr J.G. et al., 2014]
These results suggest that CTCs can be used to study tumor heterogeneity and evolution and that their analysis can be more useful than tissue profiling in detecting the tumor subclones actively involved in dissemination, since, at the primary tumor site, such clones could be diluted or be present only in a particular region.
Later studies comparing mutations and CNAs detected in single CTCs and tissue biopsies from the same patients with prostate [Lambros M.B. et al. 2018], breast [De Luca F. et al., 2016; Paoletti C. et al., 2018], colorectal [Gao Y. et al., 2017] and lung [Chemi F. et al., 2019] cancer, further confirmed the findings of Lohr and colleagues. In particular, Chemi and colleagues [Chemi F. et al., 2019] showed that, in a patient with non-small-cell lung cancer, CTCs collected at the time of surgery were molecularly more similar to the metastasis developed 10 months later (91% of overlapping mutations), than to the primary tumor (79% of overlapping mutations), whereas only 46% of primary tumor mutations were detected in CTCs, suggesting that CTCs detected at surgery derived from a subclone of the primary tumor that was the precursor of the metastasis.
Nonetheless, the aforementioned studies analyzing the molecular characteristics of single CTCs also demonstrated a high level of heterogeneity between CTCs collected from the same blood draw. In the study by Lambros and colleagues [Lambros M.B. et al. 2018], 185 CTCs collected from 14 patients with prostate cancer were analyzed for CNA profiling and showed distinct degrees of intra-patient heterogeneity that varied from patient to patient (Figure 1.14). Notably, while some patients presented very homogeneous CTCs (Figure 1.14, left side), other patients had CTCs with very different CNA profiles (Figure 1.14, right side). There were also patients presenting different CTC subtypes, such as patient 13, for whom one subset of CTCs had homogeneous CNA profiles resembling the alterations found in a metastatic tissue biopsy, a second subset presented CNAs similar to the prostatectomy biopsy, and a third subset showed very complex and non-clonal CNA patterns.
Introduction
Figure 1.14. Intra- and inter-patient genomic heterogeneity of single CTCs.
Unsupervised hierarchical clustering heat map of single CTCs collected from each patient, based on CNA profiles. Each column is a CNA profile of a single CTC, showing gains/amplifications (pink/red) and losses/deletions (light-blue/dark-blue) for the entire genome (chromosomes are indicated on the left). CTCs collected from the same patient are indicated by colors in the top row. The heat maps of each patient are organized by intra-patient diversity from left to right. [Adapted from Lambros M.B. et al., 2018]
These results further highlight the importance of performing CTC analysis at the single-cell level, since this is the only way to truly unravel intra-patient heterogeneity,
Intra-patient heterogeneity of CTCs can also change over time. Su and colleagues [Su Z. et al., 2019] monitored this by comparing CNA profiles of single CTCs collected from a patient with small-cell lung cancer at 4 different time points (before and during first-line chemotherapy, before second-line chemotherapy and during third-line chemotherapy). CTCs, which were highly homogeneous at the first time point, became progressively heterogeneous and this was possibly due to allelic losses of the initially uniform CNAs, leading to the emergence of new subclones. Overall, these studies support the use of CTCs to assess and study intra-tumor heterogeneity. However, whereas the marked diversity between CTCs on one hand carries a large amount of information about tumor evolution, on the other hand it makes the interpretation of this information much more difficult, in particular when the number of available CTCs is limited.
Other studies focused more on investigating whether the genomic profiling of single CTCs could be used to better understand therapeutic response and resistance. Dago and colleagues [Dago A.E. et al., 2014] assessed both phenotypic and genomic changes of single CTCs collected from one patient with castration-resistant prostate cancer, undergoing chemotherapy followed by androgen deprivation therapy with abiraterone. CTCs were analyzed in 4 blood samples collected before starting chemotherapy (draw 1), after progression to chemotherapy, before starting abiraterone (draw 2), after 3 weeks of treatment with abiraterone, when the patient was responding (draw 3) and after 9 weeks of treatment, during disease progression (draw 4). CTC enumeration and the evaluation of androgen receptor (AR) expression were performed using the CellSearch®. Forty-one single CTCs were isolated and analyzed for CNAs, and compared with the CNA profile of 1 bone metastasis biopsy taken at the time of diagnosis, 5 months before draw 1. By unsupervised clustering of the 41 CTCs, 3 clones were identified (Figure 1.15 A).
Introduction
Figure 1.15. Clonality and genomic alterations in single CTCs.
A) Unsupervised hierarchical clustering of CNA profiles of 1 metastatic biopsy and 41 single CTCs collected at 4 time points (draw 1-4) from a single patient with castration-resistant prostate cancer. Each column is the CNA profile of a single CTC. Gains and losses are indicated in red and blue, respectively. The colors in the top row indicate the blood sample from which the CTC was isolated (draw 1, yellow; draw 2, orange; draw 3, purple; draw 4, black). The green sample represents the metastatic biopsy. B) Plot of the AR amplification events for the 3 clones. Each line is a single CTC. [Adapted from Dago A.E. et al., 2014]
Clone A included most of the CTCs collected from draw 1 and 2, before abiraterone treatment. These CTCs were genetically similar to the metastatic biopsy and were AR-positive. Most CTCs also presented AR amplification (which was not detected in the metastasis). After starting treatment with abiraterone, when the patient was still responding (draw 3), CTCs expressed little or no AR and showed near-normal CNA profiles (clone B), suggesting that abiraterone targeted the androgen-dependent subpopulation of CTCs.
(clone C), showing again the expression of AR but distinct CNA profiles with regards to clone A. Interestingly, these CTCs presented AR amplification as CTCs in clone A. But whereas in clone A the AR amplification events were very heterogeneous among the different CTCs, in clone C all CTCs presented the same amplification profile with a higher amplification level (Figure 1.15 B), suggesting that this new abiraterone-resistant clone could represent a novel lineage, induced by selection pressure, deriving from a single resistant cell. The authors also identified, in clone C, the amplification of MYC as a possible bypass mechanism for AR-independent resistance, supporting the hypothesis that CTC genomic characterization can give hints on treatment resistance mechanisms.
The use of CTCs for detecting alterations involved in therapeutic resistance was exploited by other investigators. For instance, Paolillo and colleagues [Paolillo C. et al., 2017] investigated the presence of ESR1 mutations in single CTCs collected from patients with hormone receptor-positive metastatic breast cancer receiving endocrine therapy. In 2 out of 3 patients, the activating mutation Y537S was detected in CTCs and mirrored treatment failure. Similarly, Pailler and colleagues [Pailler E. et al., 2019] analyzed single CTCs collected from patients with anaplastic lymphoma kinase (ALK)-rearranged non-small-cell lung cancer, progressing to treatment with ALK inhibitors. By performing targeted sequencing of 48 cancer-related genes and 14 ALK mutations, the authors could detect multiple mutations in different genes involved in both ALK-dependent and ALK-independent pathways, suggesting the presence, also within the same patient, of heterogeneous subclones of tumor cells undertaking different therapeutic resistance mechanisms, that were not detectable in the corresponding tissue biopsy. Overall these results show that CTC molecular characterization can be more informative than tissue analysis for investigating therapeutic resistance mechanisms. However, they are still
Introduction A stronger support for the clinical relevance of single-CTC molecular characterization was provided by the study by Carter and colleagues [Carter L. et al., 2017]. By analyzing the CNA profiles of 88 single CTCs collected from 13 chemotherapy-naive patients with small-cell lung cancer, the authors developed a CNA-based classifier able to distinguish between patients with chemorefractory (i.e. undergoing disease progression within 3 months after the end of first-line chemotherapy) and chemosensitive disease. The classifier’s ability of identifying chemorefractory and chemosensitive patients was validated in an independent cohort of 18 patients, in which 83.3% of patients were correctly classified (Figure 1.16 A)
Figure 1.16. CTC CNA-based classifier and clinical outcome.
A) Clinical status and CTC CNA-based classification of 31 patients with small-cell lung cancer, divided in a training set (13 patients) and a testing set (18 patients). The accuracy of the CNA-based classifier is reported for the training set, the testing set and the entire cohort. B) Kaplan-Meier curves for progression-free survival and overall survival of the 31 patients according to the clinical classification (left, top and bottom) and to the CNA-based classification of patients (right, top and bottom) [Adapted from Carter L. et al., 2017]
PFS = 2.8 vs. 5.8 months, p = 0.0166), supporting the clinical validity of the classifier (Figure 1.16 B). Overall, these results demonstrate that molecular characterization of CTCs before treatment initiation can give information on tumor chemosensitivity and can potentially help predicting which patients will respond to chemotherapy.
The field of single-CTC molecular profiling is still under development and many more studies are required to prove its clinical validity. However, the results obtained so far definitely support the potential of single-CTC analysis to provide a real-time assessment of the disease, of its heterogeneity and its evolution, offering the possibility to better understand therapeutic resistance.
Scope of the thesis
Most BTC patients are diagnosed at an advanced stage, when surgery, the only potentially curative therapy, is not feasible. Chemotherapy with cisplatin and gemcitabine is the standard first-line treatment for patients with advanced biliary tract cancer, but unfortunately most patients develop resistance to this treatment and the median overall survival is less than one year. For patients with progressive disease who need to undergo a second-line therapy there is, currently, no standard treatment.
There is, therefore, a great need for new therapies in BTC. Molecular characterization data suggest that many patients carry actionable mutations and would benefit from targeted therapies. However, the use of targeted therapies is limited due to the difficulty to obtain tissue biopsies for molecular profiling.
CTCs could represent an alternative source of tumor material to perform molecular profiling, fostering personalized therapy in BTC. Moreover, CTC characterization at the single-cell level could be used to monitor the evolution and the response to treatment of the disease over time.
However, the application of CTC analysis in BTC is hindered by the fact that, by using conventional CTC detection approaches based on epithelial marker expression, CTCs are detected only in a small proportion of patients, probably due to the presence of CTCs that do not express epithelial markers. New approaches that also allow for the detection of these non-epithelial CTCs could therefore help in implementing the use of CTC analysis in BTC patients.
Based on these considerations, the ultimate goal of this study was to assess if a multilevel characterization of single CTCs could provide information on tumor molecular features and on response/resistance to treatment in patients with BTC.
Scope of the thesis Therefore, the specific aims of the thesis work were:
1) to develop a method for the detection not only of epithelial CTCs but also of those lacking epithelial markers, allowing their characterization at the single-cell level; 2) to assess if CTCs with different phenotypes have a different clinical relevance; 3) to test whether single-CTC molecular characterization could be used to perform a
molecular profiling of the tumor and to study the evolution of the disease in response to treatment.
Materials and methods
3.1.
Patient information and clinical sample collection
This was a prospective, monocentric, observational study conducted at Fondazione IRCCS Istituto Nazionale dei Tumori (Milan, Italy). For this study, 24 patients with a confirmed diagnosis of metastatic/unresectable BTC were consecutively recruited between January 2015 and March 2017. Due to the explorative nature of the study, no statistical hypothesis was postulated. The number of enrolled patients was consistent, however, with the entropy-based approach to sample size in translational clinical trials as proposed by Piantadosi and colleagues [Piantadosi S., 2005].
Patients have been treated and followed up as per clinical practice, with frequent clinical evaluations and tumor assessment with chest/abdomen CT scans and/or MRIs performed every 2-3 months. The treatment efficacy was assessed as per RECIST v1.1. Clinical information was collected from medical records and included demographic data, tumor anatomical location, tumor extension, and treatment history. The patients’ vital status was updated at the end of June 2018.
All CTC evaluations were carried out without the knowledge of the patient’s clinical status.
Samples of peripheral venous whole blood (10 mL) were drawn in EDTA tubes (K2EDTA
BD Vacutainer®, Becton Dickinson, Franklin Lakes, NJ, USA), stored at 4 °C protected from light and processed within 1 hour for CTC enrichment (the first mL of blood was discarded to avoid skin epithelial cell contamination). Blood samples were longitudinally collected at times corresponding to baseline (BL), i.e. before initiation of a new treatment line, during treatment (DT) close to clinical and imaging evaluations, at the end of treatment (EOT) and at subsequent follow-up (FU) or new treatment lines.
3.2.
Cell lines and culture conditions
The human breast cancer cell lines MCF7, MDA-MB-231 and T-47D were acquired from American Type Culture Collection and cultured in DMEM-F12 medium (Lonza, Basel, Switzerland) supplemented with 10% fetal bovine serum (FBS) (Gibco, Thermo Fisher Scientific,Waltham, MA, USA). The human dermal fibroblast cell line NHDF was acquired from Lonza and cultured in FBMTM Basal Medium (Lonza) supple
